spDarkM-93 Dataset
收藏官方服务：
资源简介：
spDarkM-93 dataset. To benchmark ACCESS and competing protein function prediction tools on functional “dark matter”, we curated a specific dataset from the Swiss-Prot dataset retaining only proteins that share <30% sequence identity and <0.5 TM-score with any previously annotated structural homolog of the same function. All proteins previously used in the training of any of the evaluated tools (ACCESS, DeepEC, CLEAN, TopEC, GraphEC) were removed. The resulting non-redundant collection, termed spDarkM-93, contains 93 proteins.
提供机构：
Zenodo创建时间：
2025-10-16
